Sends a message in the chat session and receives the model’s response. The chat maintains conversation history automatically.
Method Signatures
send_message (Non-streaming)
chat.send_message(
message: Union[list[PartUnionDict], PartUnionDict],
config: Optional[GenerateContentConfigOrDict] = None
) -> GenerateContentResponse
send_message_stream (Streaming)
chat.send_message_stream(
message: Union[list[PartUnionDict], PartUnionDict],
config: Optional[GenerateContentConfigOrDict] = None
) -> Iterator[GenerateContentResponse]
Parameters
message
string | Part | list[Part]
required
The message to send to the model.Can be:
- A simple string:
"Tell me a story"
- A Part object:
Part(text="Hello")
- A list of Parts:
[Part(text="Describe this:"), Part(inline_data=image)]
Supported part types:
- Text
- Images (PIL Image, bytes, or file data)
- Video
- Audio
- File references
Optional configuration to override the chat’s default config for this specific request.Common options:
temperature: Controls randomnessmax_output_tokens: Maximum response lengthtop_p, top_k: Sampling parametersresponse_mime_type: Output format (e.g., "application/json")
Returns
send_message
The complete model response containing:
text: The response textcandidates: List of response candidatesusage_metadata: Token usage informationmodel_version: The model version used
send_message_stream
chunks
Iterator[GenerateContentResponse]
An iterator that yields response chunks as they’re generated.Each chunk contains partial content that can be displayed incrementally.
Examples
Basic Text Message
from google import genai
client = genai.Client(api_key='your-api-key')
# Create chat
chat = client.chats.create(model='gemini-2.0-flash')
# Send message
response = chat.send_message('Tell me a story about a robot')
print(response.text)
Streaming Response
# Create chat
chat = client.chats.create(model='gemini-2.0-flash')
# Stream the response
for chunk in chat.send_message_stream('Write a poem about the ocean'):
print(chunk.text, end='')
print() # New line after streaming completes
Message with Image
from google.genai import types
from PIL import Image
# Load image
image = Image.open('photo.jpg')
# Create chat
chat = client.chats.create(model='gemini-2.0-flash')
# Send message with image
response = chat.send_message([
'What do you see in this image?',
image
])
print(response.text)
Override Configuration
# Create chat with default config
chat = client.chats.create(
model='gemini-2.0-flash',
config={'temperature': 0.5}
)
# Override config for a specific message
response = chat.send_message(
'Be creative and write a story',
config={'temperature': 1.5} # More creative for this message
)
print(response.text)
Multi-turn Conversation
# Create chat
chat = client.chats.create(model='gemini-2.0-flash')
# First turn
response = chat.send_message('What is the capital of France?')
print(f"Assistant: {response.text}")
# Second turn - chat remembers context
response = chat.send_message('What is its population?')
print(f"Assistant: {response.text}")
# Third turn
response = chat.send_message('What are some famous landmarks there?')
print(f"Assistant: {response.text}")
JSON Output
import json
chat = client.chats.create(model='gemini-2.0-flash')
response = chat.send_message(
'List 3 colors with their hex codes',
config={'response_mime_type': 'application/json'}
)
colors = json.loads(response.text)
print(colors)
Async Streaming
import asyncio
from google import genai
client = genai.Client(api_key='your-api-key')
async def stream_example():
# Create async chat
chat = client.aio.chats.create(model='gemini-2.0-flash')
# Stream response asynchronously
async for chunk in await chat.send_message_stream('Tell me a story'):
print(chunk.text, end='')
print()
asyncio.run(stream_example())
File Upload in Chat
# Upload a file first
file = client.files.upload(file='document.pdf')
# Use in chat
chat = client.chats.create(model='gemini-2.0-flash')
response = chat.send_message([
types.Part(file_data=types.FileData(file_uri=file.uri)),
'Summarize this document'
])
print(response.text)
Access Chat History
chat = client.chats.create(model='gemini-2.0-flash')
# Send some messages
chat.send_message('Hello')
chat.send_message('How are you?')
# Get comprehensive history (includes all turns)
history = chat.get_history(curated=False)
for content in history:
print(f"{content.role}: {content.parts[0].text}")
# Get curated history (only valid turns)
curated_history = chat.get_history(curated=True)
Error Handling
try:
chat = client.chats.create(model='gemini-2.0-flash')
response = chat.send_message('Hello')
print(response.text)
except Exception as e:
print(f"Error: {e}")
API Availability
These methods are available in both Gemini API and Vertex AI.